AXIOM: Self-hosted Python voice agent

Understand the project

Practical overview

Use AXIOM when you want to study or adapt a self-hosted conversational voice-agent project and can manage its Python environment and model files.

Python developers exploring conversational voice agents for robotics or edge projects.
Self-hosters interested in a browser-based voice interface connected to a local backend.

Prototype a conversational assistant that accepts speech through a browser interface.
Classify user intent and produce contextual responses using templates, semantic retrieval or an Ollama fallback.
Inspect a named speech-recognition and text-to-speech pipeline for a robotics or edge voice project.

Before you choose

The public repository displays an Apache-2.0 license.

Python 3.10 or later is required by the documented setup.
The README lists 8 GB minimum RAM and 1 GB disk for models; GPU acceleration is optional, with CPU mode supported according to the project documentation.
Kokoro and Sherpa model paths must be accessible; the README describes symlinks and environment-variable configuration.
Browser microphone permissions are documented for localhost or 127.0.0.1.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare the Python environment, configure model paths, and open the browser interface as described in the project README.

Python 3.10 or later
At least 8 GB RAM and 1 GB disk space for models
GPU acceleration is optional; the README lists 2–3.6 GB VRAM for GPU use and says CPU mode works
Kokoro and Sherpa model files accessible through the models directory or configured paths

Clone the repository, create and activate a Python virtual environment, then install dependencies with pip install -r requirements.txt.
Check the model links in models/. If they are broken, set KOKORO_PATH and SHERPA_PATH to the corresponding model locations.
From the repository, change to backend and run python main_agent_web.py.
Open http://localhost:8000 in a browser and use the microphone interface; use localhost or 127.0.0.1 for browser microphone permissions.

A locally served browser interface for voice interaction with the AXIOM backend.

The README identifies model-path setup as a first-run requirement.
The documented microphone workflow specifies localhost or 127.0.0.1 rather than an IP address.

Evidence and transparency

Source record details

Source availability: Active
Sources checked: 12 October 2026
Original website title: pheonix-delta/axiom-voice-agent · Official homepage title · retained in its source language